从局部到全局的规则模型：粒聚合研究

From Local to Global Rule-Based Models: A Study in Granular Aggregation

摘要:

顾名思义，多视图模型是从不同的视角捕捉现实界系统的模型，通常包含本地可用的特性（如属性、输入变量）。综合考虑时，必须对一群多视图模型进行聚合。当建立一个包含所有属性的整体模型不可行且不能通过合理的计算实现时，多视图模型也会出现在包含大量变量的数据中。基于模糊规则体系结构，考虑和讨论2种情形。在构建多视图模型的聚合时，一个重要的任务是为整个全局模型设置一个可靠的质量度量，使用该度量可以有效地评估由规则模型生成的单个结果的相关性。提倡用输出的信息粒来量化结果的质量，而不是一个单一的数字结果。在上述2个情形中，使用合理粒度增强原理（粒计算的基础之一）聚合了一系列多视图模型产生的结果。认为多视图模型传递的结果多样性可以通过生成结果的粒度形式进行捕获和量化。最后，讨论了相关的优化准则和优化过程。

Abstract:

Multiview models， as the name stipulates， are models capturing real-world system perceived from different points of view （perspectives）， typically engaging locally available features （attributes， input variables）. When considered together， a collection of multiview models has to be aggregated. Multiview models also arise in the presence of data with a massive number of variables when building a monolithic model involving all attributes is neither feasible nor computationally sound. In this paper， two categories of scenarios have been formulated and discussed by focusing on fuzzy rule-based architectures. An important task when building an aggregate of multiview models is to equip the overall global model with a sound measure of quality， by using which， one can efficiently assess the relevance of the individual results produced by the rule-based models. It is， therefore， advocated that the quality of the results can be quantified by an output information granule rather than a single numeric outcome. In the two scenarios outlined above， the results produced by a family of multiview models are aggregated with the use of the augmented principle of justifiable granularity-one of the fundamentals of Granular Computing. It is also advocated that the diversity of the results delivered by multiview models can be captured and quantified in the granular form of the produced result. The related optimization criterion along with the associated optimization process are discussed.